TW201819061A - Method and apparatus for abnormality diagnostic for rolling equipment - Google Patents

Abstract

An abnormality diagnostic apparatus for a rolling equipment according to the present invention is an apparatus that is connected to a data collection apparatus that collects and records process data of the rolling equipment in a time-series order and diagnoses abnormality of the rolling equipment based on the data recorded in the data collection apparatus. The abnormality diagnostic apparatus comprises at least a data extraction part, a determination part and a determination result evaluation part. The data extraction part extracts the data corresponding to the same rolled product from the data recorded in the data collection apparatus. The determination part determines whether the extracted data is within a normal range defined by a normal data group stored in a first database. The determination result evaluation part evaluates a determination result obtained by the determination part based on a rolled result of the rolled product corresponding to the extracted data. Particularly, the determination result evaluation part changes a determination standard to define the normal range when the rolled result does not match the determination result obtained by the determination part.

Description

 Method and device for abnormal diagnosis of calendering equipment  

The present invention relates to a method and apparatus for abnormal diagnosis of a calendering apparatus which is a method of calendering a metal material and manufacturing a calendered product.

In recent years, customers have become more and more strict with the specifications required for calendered products. In particular, in addition to the dimensional shape of the rolled product, the mechanical properties such as strength and ductility are converged within the allowable range, and it is important to meet the specifications required by the customer. However, in hot rolling calendering, there has not been a direct control of mechanical properties. Since the mechanical properties are closely related to the temperature history of the pressure delay, what is currently being done is to indirectly manage the mechanical properties based on the temperature information of the pressure delay.

The size or shape of the calendered product is controlled by automated equipment to maintain the required accuracy. However, in fact, the dimensional shape or temperature accuracy of the calendered product is deeply affected by the maintenance state of the equipment. In particular, when the number of personnel of the equipment maintenance staff is insufficient, there is a case where the maintenance is delayed and the quality of the rolled product is adversely affected. Therefore, the process data of the pressure delay is monitored to expect the construction of the design of the abnormal state to be automatically determined. For example, Japanese Patent No. 5158018 proposes to detect the feature quantity from the time series data of the device, and compare whether the feature quantity is similar to the past abnormal state, thereby judging the abnormal state.

In the method disclosed in the above patent documents, in order to diagnose correctly, it is necessary to store past abnormal events in advance. However, there are limits to collecting abnormal events and their characteristic quantities beforehand. Taking the hot rolling and rolling line as an example, the number of signals from equipment and sensors can reach tens of thousands. On the other hand, unusual events occur less frequently. Therefore, the collection of abnormal events and their characteristic quantities requires enormous labor. Furthermore, the collection of abnormal events and their feature quantities must be based on the definition of abnormal events, so it cannot correspond to unknown abnormal events.

 (previous technical literature)    (Patent Literature)  

Patent Document 1: Japanese Patent No. 5158018

The present invention provides a data for the abnormality diagnosis of the rolling device, which does not require pre-storing the past abnormal state for the abnormality diagnosis with high accuracy, and can also respond to an abnormality that is unknown in the abnormality diagnosis of the rolling device. Method and device.

The abnormality diagnosis method of the calendering apparatus of the present invention is a method for collecting and recording the process data of the calendering apparatus in time series by the data collecting device, and diagnosing the abnormality of the calendering apparatus according to the data recorded in the data collecting device, and having at least the following Three steps are described.

The first step is a step of extracting data corresponding to the same calendered product from the data recorded in the data collecting device. The second step is to determine whether the extracted data is in a normal range as defined by the normal data group stored in the first database. And, in the third step, the judgment result in the second step is evaluated based on the calendering result of the calendered product corresponding to the extracted material, and when the judgment result does not coincide with the calendering result, the judgment for defining the normal range is changed. Benchmark steps.

Among the above three steps, in particular, there are a first step and a second step, so that even if the data of the past abnormal situation is not stored, as long as there is normal data (that is, the data obtained when a good calendering result is obtained), Perform an abnormal diagnosis. Collecting normal data is easier than collecting abnormal data and does not require labor. In addition, as long as the abnormality diagnosis is performed based on the normal data, the undefined abnormality can be matched. Furthermore, by having the third step, the criterion for judging whether or not the normal range is updated is matched with the actual calendering result, so that the accuracy of abnormal diagnosis using normal data can be improved.

The abnormality diagnosis method of the rolling apparatus of the present invention may further include the following fourth step or fifth step. The fourth step is a step of logging the extracted data into the first database when it is judged that the extracted data is within the normal range and the rolling result of the rolled product is good. The fifth step is the step of registering the extracted data in the second database storing the abnormal data when it is judged that the extracted data is not in the normal range and the rolling result of the rolled product is not good. When the fourth step is provided, the storage amount of the normal data for determining the setting of the reference is increased, whereby the accuracy of the judgment criterion can be improved. When the fifth step is provided, the undefined abnormality may be included, and the abnormality-related data generated by the calendering device may be stored in the second database.

In the third step, it is also possible to: when the data judged to be extracted is judged to be in the normal range, but the rolling result of the rolled product is not good, the judgment criterion is changed in a strict direction. Further, in the third step, it is also possible to: when the data judged to be extracted is judged to be out of the normal range, and the rolling result of the rolled product is good, the judgment criterion is changed in the loose direction. Alternatively, it is possible to provide a sixth step of outputting an alarm to the display device when the calendered product is judged to be in a normal range despite the judgment that the extracted data is not in the normal range.

The data extracted in the first step may also be associated with a plurality of scales as components. In this case, in the second step, it is also possible to determine whether the extracted data is within a normal range based on the distance between the normal data group and the extracted data in the space of the plurality of scales. In addition, the distance between the normal data group and the extracted data can also be calculated by a plurality of scale composition methods. Furthermore, in the second step, it is also possible to use a dynamic time warping method to correct the distance between the normal data group and the extracted data.

Further, according to the present invention, there is also provided a program for causing a computer to execute the processing of each step in the abnormality diagnosis method of the above-described rolling apparatus, and a memory medium storing the program.

Further, the abnormality diagnosing device of the calendering apparatus of the present invention is a device which is connected to a data collecting device which collects and records process data of a calendering device in time series, and diagnoses an abnormality of the calendering device based on the data recorded in the data collecting device, specifically It is purchased in the following manner.

In other words, the abnormality diagnosing device of the rolling apparatus according to the present invention includes a data extracting unit, a determining unit, and a determination result evaluating unit. The data extraction department is configured to extract data corresponding to the same rolled product from the data recorded in the data collection device. The determining unit is configured to determine whether the data extracted by the data extracting unit is within a normal range defined by the normal data group stored in the first database. The determination result evaluation unit is configured to evaluate the determination result of the determination unit based on the rolling result of the rolled product corresponding to the data extracted by the data extraction unit. More specifically, the determination result evaluation unit is configured to change the criterion for determining the normal range when the determination result of the determination unit does not match the rolling result.

According to the above configuration, in particular, the data extracting unit and the judging unit are provided, whereby even if the data of the past abnormal situation is not stored, the normal data (that is, the data when the good calendering result is obtained) can be performed. Abnormal diagnosis. Collecting normal data is easier than collecting abnormal data and does not require labor. In addition, as long as the abnormality diagnosis is performed based on the normal data, the undefined abnormality can be matched. Further, by including the determination result evaluation unit, the criterion for determining whether or not the normal range is normal is updated to match the actual rolling result, so that the accuracy of the abnormality diagnosis using the normal data can be improved.

The computer may execute each process of the data extracting unit, the determining unit, and the determination result evaluating unit that constitute the abnormality diagnostic device. In other words, the computer having the memory of the program and the memory program may be configured to constitute the abnormality diagnostic device, and the program is configured to cause the computer to function as a data when the program is read from the memory by the processor. The department, the judgment unit, and the judgment result evaluation unit operate.

The determination result evaluation unit may be configured to register the extracted data in the first database when the extracted data is determined by the determination unit to be within the normal range and the rolling result of the rolled product is good. By configuring as described above, the storage amount of the normal data for determining the setting of the reference is increased, whereby the accuracy of the determination criterion can be improved.

Further, the determination result evaluation unit may be configured to: when the extracted data is determined by the determination unit to be out of the normal range, and the rolling result of the rolled product is not good, the extracted data is registered in the storage abnormal data. Second database. By constructing as described above, an unknown abnormality that is not defined may be included, and the data related to the abnormality generated by the calendering apparatus may be stored in the second database.

The determination result evaluation unit may be configured to change the criterion of judgment in a severe direction when the data to be extracted is judged to be within the normal range by the determination unit, but the rolling result of the rolled product is not good. In addition, the determination result evaluation unit may be configured to change the determination criterion toward the loose light item when the data to be extracted is determined by the determination unit to be out of the normal range, and the rolling result of the rolled product is good. Alternatively, the determination result evaluation unit may be configured to output an alarm to the display device when the data to be extracted is determined by the determination unit to be out of the normal range, and the rolling result of the rolled product is good.

The data extraction unit may also be configured to extract data based on a plurality of associated scales. In this case, the determination unit may be configured to determine whether the extracted data is within a normal range based on the distance between the normal data group in the space in which the plurality of scales are the axis portion and the extracted data. In addition, the distance between the normal data group and the extracted data can also be calculated by a plurality of scale composition methods. Furthermore, the determination unit may be configured to correct the distance between the normal data group and the extracted data using a dynamic time stretching method.

According to the present invention, even if the data of the past abnormal state is not stored, the abnormality diagnosis can be performed as long as there is normal data (that is, the data at the time of obtaining a good calendering result). Collecting normal data is easier than collecting abnormal data and does not require labor. In addition, abnormal diagnosis is performed based on normal data, thereby making it possible to correspond to an undefined unknown. Furthermore, according to the present invention, since the criterion for judging the normal range or not is updated to match the actual calendering result, the accuracy of the abnormality diagnosis using the normal data can be improved.

2‧‧‧heating furnace

3‧‧‧Hot rust remover

4‧‧‧ roughing machine

5‧‧‧Rough rolling horizontal calender

6‧‧‧Rough rolling exit side thermometer

7‧‧‧ Thermal imaging device

8‧‧‧ Rolling edge heater

10‧‧‧ Finished entrance side thermometer

11‧‧‧Fine rolling rust remover

12‧‧‧Edger

13‧‧‧ Finish rolling calender

14‧‧‧Composite meter

15‧‧‧Fine Rolling Exit Side Thermometer

16‧‧‧Output roller table

17‧‧‧ Thermal imaging device

18‧‧‧ coil entrance thermometer

19‧‧‧ coiler

20‧‧‧ Hot rolling and rolling line (calendering equipment)

22‧‧‧ data collection device

30‧‧‧Abnormal diagnostic device

31‧‧‧CPU (central processing unit)

32‧‧‧ROM (Read Only Memory)

33‧‧‧RAM (Random Access Memory, random access memory)

34‧‧‧Input and output interface

35‧‧‧System Bus

36‧‧‧Storage

37‧‧‧ Input device

38‧‧‧Display device

39‧‧‧Communication devices

101‧‧‧Information Extraction Department

102‧‧‧Determining Department

103‧‧‧Results Evaluation Department

104‧‧‧Normal paradigm database

105‧‧‧Abnormal paradigm database

Fig. 1 is a view showing a configuration example of a rolling apparatus to which an abnormality diagnostic apparatus according to an embodiment of the present invention is applied.

Fig. 2 is a block diagram showing an example of a hardware configuration of an abnormality diagnosing device according to an embodiment of the present invention.

Fig. 3 is a functional block diagram showing a part of the functions of the abnormality diagnostic apparatus according to the embodiment of the present invention.

Figure 4 is a graph showing the range of guaranteed values associated with the quality of the calendered product.

Fig. 5 is a block diagram showing the processing flow of the score calculation for abnormality determination.

Fig. 6 is a view showing an example of time series data and a calculation example of scores based on time series data.

Fig. 7 is a view showing an example of a method of determining an abnormality by a score.

Fig. 8 is a view showing a method of abnormality determination performed by the abnormality diagnostic apparatus according to the embodiment of the present invention.

Fig. 9 is a diagram for explaining the change of the determination standard.

Fig. 10 is a diagram for explaining the change of the determination standard.

Fig. 11 is a diagram showing the correspondence between the judgment result, the calendering result, and the selected processing in a table.

Fig. 12 is a flow chart showing the procedure of abnormality diagnosis performed by the abnormality diagnostic apparatus according to the embodiment of the present invention.

Embodiments of the present invention will be described with reference to the drawings. However, the embodiments shown below exemplify the apparatus or method for embodying the technical idea of the present invention, and the configuration, arrangement, processing order, and the like of the constituent members are not limited unless otherwise stated. Intention as shown below. The present invention is not limited to the embodiments described below, and various modifications can be made and carried out without departing from the spirit and scope of the invention.

 (Configuration example of rolling equipment)  

Fig. 1 is a view showing a configuration example of a rolling apparatus to which an abnormality diagnostic apparatus according to an embodiment of the present invention is applied. Here, as a typical example of the calendering apparatus, a hot rolling calendering line 20 for producing a steel sheet which is generally widely used is exemplified. The arrows in the figure indicate the flow direction of the material being calendered. The hot rolling and rolling line 20 is a line for producing a rolled product from a rolled material (hereinafter referred to as "slab").

The hot rolling and rolling line 20 is provided along the flow direction of the material to be rolled: a heating furnace 2, a hot rust removing machine (hereinafter referred to as HSB, Hot Scale Breaker) 3, a rough rolling machine 4, and a rough Rolling horizontal calender 5, edge heater 8, finishing rust remover (hereinafter also referred to as FSB, Finish Scale Breaker) 11, edger 12, finishing rolling mill 13, output out table 16, And a plurality of machines such as a down coiler 19. These devices are connected by a transfer table (not shown) and are each driven by a motor and/or a hydraulic device. Further, the hot rolling and rolling line 20 includes a rough rolling outlet side thermometer 6, a thermal image forming apparatus 7, a finish rolling inlet side thermometer 10, a multi-gauge 14, a finishing rolling side thermometer 15, and thermal imaging. A plurality of measuring devices such as the device 17, the coil inlet side thermometer 18, and the like. Hereinafter, an outline of an apparatus and a measuring device constituting the hot rolling and rolling line 20 will be described.

The heating furnace 2 is used to heat the slab. HSB3 is used to remove the oxide film formed on the surface of the slab by the heating of the heating furnace 2. The slab is mainly impacted from the nozzle by high-pressure water to remove the oxide film.

The roughing machine 4 is mainly used for pressing the rolled material in the width direction (including the state in which the slab is completed as a product, and the same applies hereinafter) in order to ensure the width accuracy. The roughing machine 4 is configured such that the roller is placed in a direction from the upper side of the production line, and the roller is placed in the width direction of the rolled material. The coarse horizontal calender 5 is composed of a single or a plurality of stands. In order to shorten the length of the production line and the rough rolling and rolling must be multi-pass, the coarse horizontal calender 5 is often constituted by a reversible calender. The coarse horizontal calender 5 is provided with a scalar descaler and is used to impinge high pressure water on the rolled material 11 belonging to the semi-product to remove the oxide film on the surface. Since the rolling is performed at a high temperature, an oxide film is easily formed, and an oxide film removing device as described above must be suitably employed. The rough rolling outlet side thermometer 6 is disposed on the outlet side of the rough rolling horizontal calender 5, and measures the surface temperature of the rolled material belonging to the half of the rolling.

The edge heater 8 is a device that raises the temperature of the end portion in the width direction of the rolled material by induction heating of electromagnetic force. The thermal imaging device 7 is disposed on the inlet side and the outlet side of the edge heater 8, respectively. Further, downstream of the flow direction of the rolled material, and on the inlet side of the finish rolling calender 13, a finish rolling inlet side thermometer 10 is disposed. The inlet side temperature of the finish rolling calender 13 is closely related to the prediction of the deformation resistance of the material. Therefore, it is necessary to measure the temperature near the processing by the finish rolling inlet side thermometer 10. Alternatively, the measured value of the rough rolling exit side thermometer 6 must be used to obtain a high-precision temperature predicted value considering the transfer time from the rough rolling exit side thermometer 6 to the finishing rolling calender 13.

FSB11 is a device for removing an oxide film on the surface of a rolled material. FSB11 is used for removing rust caused by the time that the rolled material reaches the distance from the inlet side of the finish rolling calender 13 after the rough rolling is completed, and improves the surface properties of the rolled product after the finish rolling and rolling. The edger 12 is provided on the inlet side of the finish rolling calender 13, and ensures dimensional accuracy in the width direction of the rolled product after finish rolling and rolling. Further, the edger 12 applies a plastic working by pressing to impart a temperature rise to the end portion of the rolled material in the width direction.

The finish rolling calender 13 is a tandem type calender composed of a calender in which a plurality of stands are arranged. The finish rolling calender 13 is used for calendering to obtain the accuracy of the target product of the calendered product. On the outlet side of the finish rolling calender 13, a composite gauge 14 and a finish rolling outlet side thermometer 15 are disposed. The hybrid meter 14 has a configuration in which the X-ray detectors are arranged in the width direction. Since the composite gauge 14 can measure the thickness distribution in the width direction, it is a composite type measuring device capable of measuring the thickness, the crown, the plate width, and the like. The hybrid meter 14 has a thermal imaging device inside, and measures the temperature in the width direction of the rolled product by a thermal imaging device for correcting the detected value from the X-ray. The finish rolling outlet side thermometer 15 measures the surface temperature of the rolled material after rolling. Since the temperature of the calendered product is closely related to the formation and material of the metal structure of the product, it must be managed at an appropriate temperature.

The output roller table 16 is a device for cooling the calendered product by cooling water in order to control the temperature of the calendered product. Further, instead of the general output roller cooling device, a forced cooling device may be provided or provided. In order to prevent the temperature of the end portion of the rolled product from decreasing in the width direction, the output roller path 16 may be applied with an edge mask to which no cooling water is applied in the width direction. A thermal imaging device 17 and a coil inlet side thermometer 18 are disposed on the outlet side of the output roller 16 and on the inlet side of the coiler 19. The coil inlet side thermometer 18 measures the surface temperature of the calendered product after calendering. Since the temperature of the calendered product is closely related to the formation and material of the metal structure of the product, it must be managed at an appropriate temperature. The coiler 19 is a device for coiling a product for conveying a rolled product. The roll-shaped rolled product coiled by the coiler 19 is called a coil.

The hot rolling and rolling line 20 is provided with a data collecting device 22. The data collection device 22 collects continuously or intermittently: set values or actual values of the devices constituting the hot rolling and rolling line 20, measured values measured by the measuring device, and even used to properly operate the device. The amount of operation, etc., is recorded on a recording device such as a hard disk. For example, the rolled product taken up by the coiler 19 is a plate thickness, a plate width, and the like measured by the composite gauge 14, the finish rolling outlet side thermometer 15, the coil inlet side thermometer 18, and the like. The quality indicators such as the crown height, the temperature at the exit side of the finishing roll, and the temperature at the inlet side of the coil determine the guaranteed value to the client. The data collection device 22 collects and records process data containing the above-described quality indicators as components in time series. In addition, the data collection device 22 may be configured by a single computer or may be connected to a plurality of computers on the network.

 (Hardware configuration example of abnormality diagnostic device)  

Fig. 2 is a block diagram showing an example of the hardware configuration of the abnormality diagnostic device 30 of the present embodiment. The abnormality diagnostic device 30 is a computer including a CPU (central processing unit) 31, a ROM (Read Only Memory) 32, and a RAM (Random Access Memory) 33. The input/output interface 34, the system bus 35, the storage 36, the input device 37, the display device 38, and the communication device 39.

The CPU 31 is a processing device that executes various arithmetic processing using a program or data stored in the ROM 32 or the RAM 33. The ROM 32 is a memory device for reading a basic program or an environmental file for causing a computer to realize each function. The RAM 33 is a main memory device that stores data necessary for execution of the program executed by the CPU 31 or each program, and can be read and written at high speed. The input/output interface 34 is a device for interconnecting various hardware and system bus bars 35. The system bus 35 is a communication path shared by the CPU 31, the ROM 32, the RAM 33, and the input/output interface 34.

Further, the input/output interface 34 is connected to hardware such as the input device 37, the display device 38, the memory 36, and the communication device 39. Input device 37 is a device that processes input from a user. Display device 38 is a device that displays information about diagnostic results and/or abnormal diagnostics. The storage device 36 is a large-capacity auxiliary memory device for storing programs or data, such as a hard disk device or a non-volatile semiconductor memory. The communication device 39 is a device that can communicate with the data collection device 22 by wire or wirelessly.

 (The function of the abnormality diagnostic device)  

Fig. 3 is a functional block diagram showing a part of the functions of the abnormality diagnostic device 30 of the present embodiment. The abnormality diagnostic device 30 includes a data extracting unit 101, a determining unit 102, and a determination result evaluating unit 103. The program read from the ROM 32 (see FIG. 2) of the abnormality diagnostic device 30 is executed by the CPU 31 (see FIG. 2), whereby the functions of the above-described functional units 101, 102, and 103 are realized in the computer, in other words, in the computer. The function as the abnormality diagnostic device 30 is realized. Further, the abnormality diagnostic device 30 is provided with a normal pattern database 104 belonging to the first database and an abnormal paradigm database 105 belonging to the second database. The normal paradigm database 104 and the abnormal paradigm database 105 are constructed in the storage 36 (see FIG. 2). Further, a program for causing a computer to function as the abnormality diagnostic device 30 is provided through a memory medium (for example, a CD-ROM, a DVD, a USB memory, or the like) that can be read by a network or a computer.

The data extracting unit 101 is configured to take out the data recorded in the data collecting device 22 in units of coils. The data collection device 22 records the data of each item in time series. The data extracting unit 101 extracts data corresponding to the same coil from the data recorded in the data collecting device 22 based on the time information of the data. Further, the data extracted by the data extracting unit 101 is composed of a plurality of scales associated with a thickness and a load.

The determination unit 102 is configured to determine whether or not the data extracted by the data extraction unit 101 is within the normal range. Here, Fig. 4 is a view showing the data extracted by the data extracting unit 101 and the tolerance of the guaranteed value set for the extracted data. For calendered products, tolerances with guaranteed values have been set for each quality indicator. The pressure delay is to control each device in such a way that each quality indicator is within its tolerance. However, there are cases in which the quality index is out of tolerance due to various reasons such as inflexibility of the device that is not sufficiently maintained, improper control gain, and the like. The normal range for the determination by the determination unit 102 is associated with the tolerance of the guarantee value of the quality indicator.

The determination unit 102 is configured to calculate a score from the data extracted by the data extraction unit 101. Here, FIG. 5 is a block diagram showing a processing flow of the score calculation for abnormality determination. First, the data obtained so far, the learning material produces a distribution model. The model can also be a probability model or a statistical model. Next, a learned model is used to score the degree of abnormality of the data or the degree of abnormality of the model. Fig. 6 is a view showing an example of the time series data extracted by the data extracting unit 101 and a calculation example of the score based on the time series data.

The method of determining the abnormality from the calculated score is generally a determination based on the threshold. Fig. 7 is a view showing an example of a method of determining an abnormality by a score. In Figure 7, the line drawn by the dashed line shows the threshold for the score. However, the accuracy of the abnormality determination of the method depends on the determination method of the threshold, so the method can be said to be difficult to fully guarantee the accuracy of the abnormality determination.

The determination unit 102 is configured not to judge by a single scale, but to combine the determinants that can be determined from the attributes to determine the abnormality. Fig. 8 is a view showing a method of abnormality determination by the determination unit 102. Here, in the plane defined by the scale 1 and the scale 2, dots defined by a combination of the respective scores (scores representing values) of the scale 1 and the scale 2 are drawn. In Fig. 8, the data groups grouped as "normal data groups" are stored in the data group of the normal paradigm database 104. The normal paradigm database 104 stores only the data extracted by the data extracting unit 101 so far, and the judgment result evaluating unit 103 described below finally determines that it is normal normal data.

The determining unit 102 measures the data on the plane defined by the scale 1 and the scale 2 by a multidimensional scale composition method and/or a k-nearest neighbor algorithm, or a density ratio estimation method or the like. The distance (i.e., the distance between the data extracted by the data extracting unit 101 and the normal data group). Furthermore, the dynamic time stretching method can also be used to correct the distance. In Fig. 8, the curve drawn by the broken line defines the judgment line of the normal range, and sets the normal data group stored in the normal paradigm database 104 as a reference. The determination unit 102 determines that the data that is closer to the normal data group than the determination reference line is within the normal range, and determines that the distance from the normal data group is farther than the determination reference line is outside the normal range. The gauge 1 is, for example, a plate thickness, and the gauge 2 is preferably a load that is in close contact with the thickness of the plate. In the case where the thickness of the plate is abnormal, since the abnormality is also likely to occur in a load in a relationship with the thickness of the plate, the scale 1 and the ruler 2 provided in the above manner can be easily excluded. The noise of the sensor can be clearly judged abnormally.

The determination result evaluation unit 103 is configured to evaluate the determination result by the determination unit 102. The criterion for the evaluation used by the determination result evaluation unit 103 is the data extracted by the data extraction unit 101, that is, the actual calendering of the rolled product corresponding to the data of the object within the determination range determined by the determination unit 102. result. Specifically, the calendering result is a result of whether or not each is within the quality control value for the thickness, the plate width, the temperature, the shape, and the like after the end of the rolling. The confirmation of the rolling result is automatically performed by a dedicated device or the like, and is preferably obtained online via the communication device 39. However, the result confirmed by the person can also be input through the input device 37.

In detail, the determination result evaluation unit 103 compares the determination result by the determination unit 102 with the rolling result as the quality criterion, and evaluates whether or not the determination result matches the rolling result. Further, processing corresponding to the evaluation result is performed. Here, Fig. 11 is a table showing the correspondence between the judgment result, the calendering result, and the selected processing in a table. Hereinafter, what kind of processing is performed in accordance with the evaluation result will be described with reference to Fig. 11 .

First, the judgment results within the normal range are in good agreement with the good calendering results. In this case, the determination result evaluation unit 103 registers the data that is determined to be within the normal range by the determination unit 102 in the normal paradigm database 104. By the above method, the storage amount of the normal data for judging the setting of the reference is increased, whereby the accuracy of the judgment criterion can be improved. In addition, the data registered here is the data corresponding to the same calendered product (coil) drawn by the data extracting unit 101, and each piece of data is measured in the length direction of the coil (quality The difference model of the indicator) is different. The normal paradigm database 104 stores normal data with various paradigms.

On the other hand, the judgment result within the normal range does not coincide with the result of the poor calendering. In this case, the determination result evaluation unit 103 changes the determination criterion of the determination unit 102 in order to increase the accuracy of the determination after the next time. Specifically, the data that is judged to be within the normal range is judged to be outside the normal range, and the criterion for determination is changed in a severe direction. As an example, in Fig. 9, in the plane defined by the scale 1 and the scale 2, the data A is drawn. This data A is located on the same side as the normal data group with respect to the curve showing the judgment criterion 1. Therefore, according to the judgment criterion 1 shown in Fig. 9, the data A is judged to belong to the normal range. However, if the actual calendering result of the calendered product corresponding to the data A is a bad situation, the judgment that the data A is within the normal range cannot be said to be correct. Therefore, in this case, as shown in FIG. 10, the change from the determination criterion 1 to the determination criterion 2 is performed. According to the judgment criterion 2, the data A is judged to be outside the normal range, and thus coincides with the actual calendering result.

The judgment result that is not within the normal range is consistent with the result of the bad calendering. In this case, the determination result evaluation unit 103 registers the data that is determined to be outside the normal range by the determination unit 102 in the abnormal paradigm database 105. In the above manner, an unknown abnormality that is not defined may be included, and the abnormality-related data generated in the hot rolling and rolling line 20 is stored in the abnormal paradigm database 105. In addition, the paradigm of the newly registered data, specifically, the difference paradigm of the scale (quality index) for the length direction of the coil and the abnormal data group that has been registered in the abnormal paradigm database 105 The paradigm is compared to determine if an undefined unknown exception has occurred.

On the other hand, it is outside the normal range and does not match the good calendering results. In this case, the determination result evaluation unit 103 changes the determination criterion of the determination unit 102 in order to increase the accuracy of the determination after the next time. Specifically, the data that is judged to be outside the normal range is judged to be within the normal range, and the criterion for determination is changed in a loose direction. However, if the calendering result is good or the data is outside the normal range, there will be doubts about the potential loss of the device. Therefore, in this case, the operator investigates the absence of the device and can also output an alarm screen on the display device 38.

 (order of abnormal diagnosis)  

According to the abnormality diagnostic device 30 having the function described above, the abnormality diagnosis of the rolling apparatus is performed in the following order. Hereinafter, the procedure of the abnormality diagnosis by the abnormality diagnostic device 30 will be described using the flowchart shown in FIG.

In step S1, the abnormality diagnosing device 30 extracts data corresponding to the same rolled product from the data recorded in the data collecting device 22.

Next, in step S2, the abnormality diagnostic device 30 determines whether the data extracted in step S1 is within the normal range defined by the normal data group stored in the normal paradigm database 104. In detail, the abnormality diagnostic device 30 measures the distance between the normal data group and the data extracted in step S1 by a plurality of scale formation methods or the like. In this case, it is preferable to correct the distance by using a dynamic time warping method (DTW). And, based on the measured distance, it is judged whether or not the data extracted in step S1 is within the normal range.

Next, the abnormality diagnostic device 30 evaluates the result of the determination in step S2 based on the rolling result of the rolled product corresponding to the data extracted in step S1. In detail, when it is determined in step S2 that the data extracted in step S1 is within the normal range, the abnormality diagnostic device 30 determines in step S3 whether the result of the affirmative determination in step S2 is the actual calendering result. Match.

When the affirmative judgment result in the step 2 coincides with the actual calendering result, that is, the actual calendering result is in a good condition, the abnormality diagnostic device 30 selects the process of step S4. In step S4, the abnormality diagnostic device 30 registers the data extracted in step S1 in the normal paradigm database 104.

When the result of the affirmative determination in step S2 does not coincide with the actual calendering result, that is, the actual calendering result does not belong to a good condition, the abnormality diagnosing device 30 selects the processing of step S5, and in step S5, the abnormality diagnosis The device 30 changes the criterion for determining the determination to be performed in step S2 in a severe direction, and registers the data extracted in step S1 in the abnormal paradigm database 105.

Further, when it is determined in step S2 that the data extracted in step S1 is not within the normal range, the abnormality diagnostic device 30 determines in step S6 whether the negative determination result in step S2 coincides with the actual calendering result. .

When the negative judgment result in step S2 coincides with the actual calendering result, that is, when the actual calendering result is not good, the abnormality diagnosing device 30 selects the processing of step S7. In step S7, the abnormality diagnostic device 30 registers the data extracted in step S1 in the abnormal paradigm database 105.

When the negative determination result in step S2 does not coincide with the actual rolling result, that is, when the actual rolling result is good, the abnormality diagnostic device 30 selects the processing of step S8. In step S8, the abnormality diagnostic device 30 changes the determination criterion to be determined in step S2 in the loose direction. Alternatively, the abnormality diagnostic device 30 performs an output of an alarm to the display device 38.

According to the abnormality diagnosis performed in the above-described order, the normal paradigm database 104 and the abnormal paradigm database 105 are flexibly used in order to determine the judgment criteria of the abnormal coil and the normal coil in accordance with the rolling state in response to the daily change. . In this way, by combining the normal paradigm database 104 and the abnormal paradigm database 105, the accuracy of the abnormality detection side can be improved, and by comparison with the normal data group stored in the normal paradigm database 104. So that it can correspond to an unknown exception.

Claims (16)

 An abnormality diagnosis method for a calendering apparatus is a method for collecting and recording process data of a calendering apparatus in time series by a data collecting device, and diagnosing an abnormality of the calendering apparatus according to data recorded in the data collecting apparatus, the calendering apparatus The abnormal diagnosis method includes: a first step of extracting data corresponding to the same rolled product from data recorded in the data collecting device; and determining whether the extracted data is in accordance with a normal data group stored in the first database a second step in the normal range defined; and evaluating the judgment result in the second step according to the calendering result of the calendered product corresponding to the extracted material, when the judgment result does not coincide with the calendering result The third step of defining the criterion for determining the aforementioned normal range is changed.    The abnormality diagnosis method of the rolling apparatus according to the first aspect of the invention is further characterized in that, when it is judged that the extracted data is within the normal range, and the rolling result of the rolled product is good, the aforementioned extraction is performed. The data is registered in the fourth step of the aforementioned first database.    The abnormality diagnosis method of the rolling apparatus according to the first or the second aspect of the patent application is further provided when: The fifth step of registering the extracted data in the second database storing the abnormal data.    The method for abnormality diagnosis of a rolling apparatus according to any one of the preceding claims, wherein in the third step, when it is determined that the extracted data is determined to be within the normal range, the foregoing When the rolling result of the rolled product is not good, the above criteria are changed in a severe direction.    The abnormality diagnosis method of the calendering apparatus according to any one of the items 1 to 4, wherein in the third step, when it is judged that the extracted data is judged to be not within the aforementioned normal range, When the rolling result of the rolled product is good, the judgment criterion is changed in the loose direction.    The method for abnormally diagnosing a rolling apparatus according to any one of the first to fourth aspects of the present invention, wherein the calendering result of the calendered product belongs to the following, although it is judged that the extracted material is not within the normal range When it is good, the sixth step of outputting an alarm to the display device.    The method for abnormally diagnosing a rolling apparatus according to any one of the preceding claims, wherein the data extracted in the first step is data based on an associated plurality of scales. In the second step, measuring the distance between the normal data group and the extracted data in the space with the plurality of scales as the axis, and determining whether the extracted data is in the foregoing according to the distance Within the normal range.    The abnormality diagnosis method of the rolling apparatus according to claim 7, wherein in the second step, the dynamic time stretching method is used to correct the distance.    An abnormality diagnosing device for a calendering device is a device that is connected to a data collecting device that collects and records process data of a calendering device in time series, and diagnoses an abnormality of the calendering device according to data recorded in the data collecting device, the calendering device The abnormality diagnosing device includes: a data extracting unit configured to extract data corresponding to the same rolled product from data recorded in the data collecting device; and a determining unit configured to determine whether the extracted data is in accordance with Stored in a normal range defined by the normal data group of the first database; and the judgment result evaluation unit is configured to evaluate the judgment result of the judging unit based on the calendering result of the calendered product corresponding to the extracted data; The determination result evaluation unit is configured to change the criterion for determining the normal range when the determination result of the determination unit does not match the rolling result.    The abnormality diagnosing device for a rolling device according to claim 9, wherein the determination result evaluation unit is configured to determine that the extracted data is determined by the determination unit to be within the normal range, and the calendered product is When the calendering result is good, the extracted data is registered in the first database.    The abnormality diagnosing device of the rolling device according to the ninth or tenth aspect, wherein the determination result evaluation unit is configured to determine that the extracted data is not within the normal range by the determination unit. And if the calendering result of the calendered product is not good, the extracted data is registered in the second database storing the abnormal data.    The abnormality diagnosing device of the rolling apparatus according to any one of the items of the present invention, wherein the determination result evaluation unit is configured to determine that the data to be extracted is determined by the determination unit In the normal range, if the rolling result of the rolled product is not good, the above criteria are changed in a severe direction.    The abnormality diagnosing device for a rolling apparatus according to any one of the above-mentioned claims, wherein the determination result evaluation unit is configured to determine that the extracted data is not in the judgment by the determination unit. In the normal range, if the rolling result of the rolled product is good, the judgment criterion is changed in the loose direction.    The abnormality diagnosing device for a rolling apparatus according to any one of the above-mentioned claims, wherein the determination result evaluation unit is configured to determine that the extracted data is not in the judgment by the determination unit. In the above normal range, when the rolling result of the rolled product is good, an alarm is output to the display device.    The abnormality diagnosing device for a rolling apparatus according to any one of claims 9 to 14, wherein the data extracting unit is configured to extract data having a plurality of associated scales as components, and the determining unit The method is configured to: measure a distance between the normal data group and the extracted data in a space with the plurality of scales as a shaft portion, and determine, according to the distance, whether the extracted data is within the normal range .    The abnormality diagnosing device for a rolling apparatus according to claim 15, wherein the determining unit is configured to correct the distance by using a dynamic time stretching method.